Decision-Aligned Evaluation of Uncertainty Quantification
Annika Schneider, Tommy Rochussen, Joshua Stiller, Vincent Fortuin
cs.LG, cs.AI, stat.ML
Submitted: 2026-06-25
Code: https://github.com/fortuinlab/prior-weighted-utilities
License: http://creativecommons.org/licenses/by/4.0/
The gist: Uncertainty estimates in machine learning are typically evaluated using generic metrics such as the negative log-likelihood and expected calibration error, yet good performance on such metrics does
Terminology
Abstract
Uncertainty estimates in machine learning are typically evaluated using generic metrics such as the negative log-likelihood and expected calibration error, yet good performance on such metrics does not necessarily imply high utility in downstream decisions. We introduce decision-alignment, a criterion that reveals which evaluation metrics meaningfully align with downstream utilities. Applying this framework, we show that many widely used uncertainty metrics are either misaligned with common decision problems or encode pathological prior beliefs about the downstream task. We then propose prior-weighted utility metrics, a special class of proper scoring rules that provides decision-aligned uncertainty evaluation. Across benchmark experiments and real-world case studies, our metrics consistently align with realized decision utility, while conventional metrics do not. Our results surface flaws in the current UQ evaluation protocol and offer a principled extension of existing metrics toward decision-relevant UQ evaluation.
Sources
- Patterns, predictions, and actions: A story about machine learning
- Uncertainty Baselines: Benchmarks for Uncertainty & Robustness in Deep Learning
- Calibration Error for Decision Making
- Truthfulness of Decision-Theoretic Calibration Measures
- Scalable Utility-Aware Multiclass Calibration
- Quality Metrics in Recommender Systems: Do We Calculate Metrics Consistently?
- A no-regret generalization of hierarchical softmax to extreme multi-label classification
- Epistemic Artificial Intelligence is Essential for Machine Learning Models to Truly 'Know When They Do Not Know'
- Aligning the Evaluation of Probabilistic Predictions with Downstream Value
- Evaluation Metrics as Averaged Outcomes of Fair Gambles
- A Consequentialist Critique of Binary Classification Evaluation: Theory, Practice, and Tools
- Towards Clear Expectations for Uncertainty Estimation
Related papers
- Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
- AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
- Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
- Convergence issues in Relational Concept Analysis based on AOC-posets
- Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
- Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks